Solutions

Find the questions that let your test down

Paste the answer key and the candidates' responses, and see which items are too easy, too hard or fail to discriminate, and how reliable the whole test is. Free, with no account.

What the analysis shows

Difficulty and discrimination

The share of candidates answering each item correctly, and how well it separates strong candidates from weak ones.

Item–total correlation

Point-biserial correlation for every item, to spot questions that work against the rest of the test.

Reliability

Cronbach's alpha for the whole test, with the thresholds ExamQR recommends.

Distractors

How often each wrong option was chosen, so you can rewrite the options nobody picks.

IRT models

Rasch (1PL), 2PL and 3PL: difficulty, discrimination and guessing parameters, information functions and candidate ability.

Reports to keep

Export the full report to Word or PDF, and the item and candidate tables to Word and Excel.

Three steps

Prepare

One line with the answer key and one line per candidate, from Notepad or Excel. Sample data is included.

Analyze

Choose the IRT model and start. 5 to 200 questions, 10 to 5,000 candidates.

Improve

Fix or drop weak items and export the report for your team.

Built into ExamQR exams

For exams you run on ExamQR, item statistics are part of the exam's reports, with nothing to export. The free tool works with results from any source: paper answer sheets, another platform or your own scanner. Data stays in your browser session for 2 hours and is never saved to an account.

New to item statistics? Read how CTT and IRT work and what the thresholds mean

Run your next exam on ExamQR

Online exams with automatic grading and statistics for every question.